Abstract. Accuracy of current automatic image labeling methods is under the requirements of annotation-based image retrieval systems. The performance of most of these labeling meth...
Invariant representations in object recognition systems are generally obtained by pooling feature vectors over spatially local neighborhoods. But pooling is not local in the featu...
Y-Lan Boureau, Nicolas Le Roux, francis bach, Jean...
Despite advances in the archiving of digital video, we are still unable to efficiently search and retrieve the portions that interest us. Video indexing by shot segmentation has b...
Sameer Antani, David J. Crandall, Rangachar Kastur...
Describing shots through the occurrence of semantic concepts is the first step towards modeling the content of a video semantically. An important challenge is to automatically se...
In this paper, we present the FOX-MIIRE 3D-Model Search Engine. Our search engine is based on Adaptive Views Vlustering (AVC) algorithm [4]. The AVC method uses statistical model ...